📊 FinAgent

This repository accompanies the tutorial article
RAG Agent with Custom Tool: Development to Deployment
on Medium

FinAgent is a retrieval-augmented generation (RAG) financial insights agent that delivers real-time, context-aware explanations of market trends and financial concepts.
It leverages LangChain, FAISS, LangGraph, and GPT-4o, storing conversation history in PostgreSQL for persistent context.

Response

Query: "Explain NVDA trend from 01/01/2024 to 01/01/2025"

---

✨ Features

• 🧠 RAG pipeline: Uses LangChain + FAISS for retrieval + LangGraph orchestration to deliver detailed, explainable answers.
• 🔗 FastAPI service: Exposes an HTTP GET / endpoint for real-time querying.
• 🗄 Persistent history: Saves conversation state to PostgreSQL using psycopg_pool and PostgresSaver.
• ⚙️ Scalable deployment: Containerized with Docker, deployable via Kubernetes.

---

⚙️ Tech Stack

• LangChain, FAISS, LangGraph, GPT-4o
• FastAPI
• PostgreSQL
• Docker & Kubernetes

---

🚀 Getting Started

Prerequisites

• Python 3.9+
• Docker
• Kubernetes (e.g., Minikube)
• PostgreSQL instance

---

🛠 Installation & Setup

1. Clone the repository:

git clone https://github.com/raoashish10/FinAgent.git
cd finagent

2. Create .env file:

touch .env

Example .env:

DB_URI=postgresql://username:password@host:port/dbname
OPENAI_API_KEY=your_openai_key

3. Install dependencies:

pip install -r requirements.txt

---

🧪 Usage

The API is served by default at localhost:8000.

✅ curl example:

curl -X 'GET'   'http://127.0.0.1:8000/?query=Explain%20NVDA%20trend%20from%2001%2F01%2F2024%20to%2001%2F01%2F2025'   -H 'accept: application/json'

📄 Example response (as seen in FastAPI docs):

{
  "response": "NVIDIA's stock trend between January 2024 and January 2025 shows..."
}

Deploy to Kubernetes

minikube start
kubectl apply -f kubernetes/postgres-deployment.yaml
kubectl apply -f kubernetes/finagent-api-deployment.yaml